There are several VW platforms to develop an interactive virtual environment, Second Life,
Active Worlds and Wonderland, to name a few. All of them consist of similar components
such as avatars, buildings, scripting components and built-in features. We chose
Wonderland because it was conceived to work with 3D standards, it is open-source and
multi-platform (java-based). We have developed our prototype in WL 0.4 where 3D content
is represented in X3D standard format. WL 0.5 works with COLLADA, a well-extended 3D
interchange format. We are now migrating to version 0.5 available only for developers.
Once selected the VW platform, we studied how to incorporate iObjects in WL and how to
capture an event (i.e. an object interaction) in WL and communicate it to the external and
generic iObject manager. In particular, our prototype presents results obtained using an
iObjectDoor. In WL, doors are merely holes allowing to pass through them to avatars and so
change from one room to another one. An iObjectDoor adds an additional nuance letting
pass through it only avatars having permission, that is, avatars who comply with the norms
established by the multiagent system refered in Figure 4. Figure 5 shows two simulations
exploiting norm compliance for an iObjectDoor in Wonderland. Client 1 (named c1a) sees the
door in green because he complies with the norm allowing to enter the next room. Client 2
(named c2) sees it in red because he does not comply the norm. Note that both snapshots
correspond to the same door in the same virtual world but thanks to a multi-view scheme
both clients see the same door with different colors depending on their permission to pass
through it. Avatars without access permission have always the collision control enabled so
|
|
that they can never get closer to the door.
Fig. 5. Controlling norm compliance by means of an iObjectDoor in Wonderland (on the left:
Client1, on the right: Client2)
Virtual worlds can be exploited as dynamic information spaces, for example, adapting the
visualization of a virtual object depending on the participant profile or previous activities.
As mentioned before, we propose an iObject multi-view scheme by keeping different 3D
models of the iObject. All clients share an indexed set of visual representations (red door,
green door, glazing door, etc.), but only one is active for each client in a given moment.
Figure 6 shows the multi-view scheme of an iObjectDoor in Wonderland. On the left side,
Client 2 sees glazed red door because he has permission to see the next room but not to pass
through it. On the right side, Client 3 has both permission to see and to pass through it.
Controlling and Assisting Activities in Social Virtual Worlds 23
Fig. 6. Snapshots showing multi-view scheme. Client 2 and client 3 views of the iObjectdoor
on left and right pictures, respectively
When an avatar is near to the door and clicks the mouse over the door, the iObjectDoor
captures the event and asks the iObjects manager whether the client complies norms
allowing to access the room (e.g.. in an auction, the buyer has paid registration fee). Then, if
the answer is affirmative, the iObjectDoor runs the local animation and notifies it to the
server so that the rest of clients also visualize it.